Most of the time I use a Linux distro in my homelab, like for my storage nodes. When you want to download a recent .ISO file, you can choose one of the many mirrors that are available. In the list with mirrors you see other recognizable brands like Leaseweb, TransIP and a lot of other hosting/infra related firms (and universities).
Maybe StorJ could also benefit from being a mirror and gets some brand recognition in return from inside and outside the opensource community? It doesn’t need much more then hosting a static website with a storage backend and bandwidth. I know this isn’t free for StorJ, that’s why I said this could be something for the long run.
For example, Ubuntu releases is currently ±70GB (yes GB not TB). The complete package archive is somewhere in the neighborhood of ±4TB. I’m still trying to figure out how much upload/egress traffic this could generate. Aachen University (Germany) share some network traffic numbers, but they are hosting a lot of mirrors. https://ftp.halifax.rwth-aachen.de/
But there are other much smaller repositories out there, like Linux Mint with a total of 200GB storage. With Linux Mint is an plus side, you can put StorJ on category World and be on the top of the list. Cant get a better position in the list then in the top, right? https://linuxmint.com/mirrors.php
I’m currently setting things up for trying it out as a little proof of concept with my own gateway-st. I’m still in the proces of the initial sync, after that I have to upload everything to StorJ (the initial sync is slow). So the idea can still fail, but in theory it should work and could also be a great showcase about the performance of a distributed network.
Correct, used to run a big one with a lot of Linux distros. But was never impressed by the throughput . Think my ISP or something else was throttling torrent traffic. But still think it could help for the brand recognition, you don’t need a marketing team for this
By the way, gateway st isn’t gonna work for now. Directory listing is apparently not possible. Gateway Mt does support it sort of. But not in a way that’s gonna work (I think). For now I have to put a pin into it, other things to do.
Question too: how safe is minio gateway that gateway st uses? It seems that minio stopped developing it a few years ago and pulled the plug last year. Or is this another product? Maybe @Alexey knows?
Three people who download isos from http or ftp may or may not see it and those people are not decision makers anywhere.
Personally the url from where to download anything does not even register in my mind. And as Toyoo mentioned stuff like iso images — fixed large blobs of data — BitTorrent exists.
We forked minio to our gateway and supporting this fork inside our gateways.
Could you please elaborate about this?
I know that if you use a self managed encryption, then without a special option to disable keys (paths) encryption, you must use trailing / for prefixes, otherwise you will quickly hit the limit for the listing. This is related to how objects are processed to give you an ordered list: if you use unlimited prefixes (without a trailing /), the gateway is forced to decrypt all objects (because prefixes are parts of the object key), then sort them and then provide to you. To do not exhaust resources on the gateway we have a limit how much such objects can be returned.
But you can create a project with Storj-managed encryption, it creates access keys with disabled paths encryption by default exactly to get ordered listing and don’t hit the listing limit. Or you can create a new access grant with the disabled paths encryption, using an API key and the encryption phrase from the CLI, e.g.
But does it still get security patches? Is it safe to host a site and publish it to the outside world where hackers and AI agents are trying to infiltrate everything? Latest version in github is from Jul 29, 2025. Last few months a lot of other products are releasing more and more security patches (with the help of AI), trying to keep things safe.
It works likes documented in the documentation, it looks for an index.html or it shows the content of the folder in xml style. IYou cant like just see a folder structure like this: https://pub.linuxmint.io/
I hoped that there would be a config flag to change this, but could not find one.
The hosted gateway have a directory listing, but its a very fancy one. With a separate download button for downloading a file and when you click the file itself it want to give you a preview of the file and its placement around the world instead of downloading it.
For mirrors to work, it needs something very basic. Like the example url I gave, but I can give it an other try if you think it will work with the instructions you provided.
I did try this with a new project with gateway st (not the hosted gateway). But hit a wall when I try to use an new access grant instead of the api key. The gateway st setup keeps giving errors, something about a different macaroon (dindt save the exact error). Eventually I switches back to self managed keys and created an api key. The api key works immediately.
Dont know if the gateway-st manual (including requirements section) is lacking, if it is a bug or that I’m just doing something wrong. But I gave up on the storj managed encryption
Of course, as everything else in our repo, you may check the history on our repository.
It doesn’t. It’s our satellite UI and the linksharing services, which can show prefixes as “folders” and “files”.
You can run a gateway-st to show objects, but it will not be a publicly available service, they will need to provide an access key and the secret key on the authorization page to have an objects browser.
To have a public access with an object browser you need to host a linkshare service instead. Or generate a public link on our linkshare: Creating Public Buckets and Embedding Content - Storj Docs
Or run your gateway-st in a static website mode. Or use a simple website which can show the mounted bucket, which you mount with Object Mount or rclone mount. There are many options how to do that
You shouldn’t need to do this. Directory listings in HTTP were never truly defined, they’re just for human convenience. As such, you could just as well upload an index.html manually and use it as the entry point for humans.
I think you’re asking if the original, open-source Minio repo is getting patches, and the answer is no: the project has been archived. The min.io company is now marketing AIStor, which is “Built for Enterprise AI Workloads” according to the website. It’s no longer open source.
